MVSFORUMS.com

[hsubra]

Hi,

I need pickup block from my report. For example

ABC, SYSTEMS
REPORT Q
-----------------------------
AEXREC001 1222.22
---------002 0000.00
---------003 0000.50
*****------1222.72
BDATA 001 ...
..
******
AEXREC 001
...
******
Here I need all the records start from AEXREC to endwith next ****** in position 1 to 6. ( I kept the block in red) .. Is it possible do with ICETOOL? Is there anyway i can use expressions in STARTREC & ENDREC ?

Please Advise and appriciate your immediate help ...

Thanks You,
Subra

[Frank Yaeger]

No, you can't use a logical condition in STARTREC or ENDREC. But you can do what you want with the following DFSORT/ICETOOL job. (I assumed you really had ****** in your input record rather than the ***** you showed.)

[hsubra]

Hi Frank,

Thanks for the immediate help. Your JCL work fine , if i have one block to select. I have multiple blocks line that .

My input is

ABC, SYSTEMS
REPORT Q
-----------------------------
AEXREC001 1222.22
---------002 0000.00
---------003 0000.50
******------1222.72
BDATA 001 ...
******---
AEXREC00X 1222.22
---------00Y 0000.00
---------00Z 0000.50
******------1222.72
BDATA 001 ...
******---

My output should be ,
AEXREC001 1222.22
---------002 0000.00
---------003 0000.50
******------1222.72
AEXREC00X 1222.22
---------00Y 0000.00
---------00Z 0000.50
******------1222.72

Thanks
Subra

[Frank Yaeger]

I can't think of a solution for extracting multiple blocks other than to use an E15 exit with the appropriate logic.
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

[Alain Benveniste]

Here is the Jcl that do what you want. you also need the new DFSORT PTF from Dec 2004.

[Frank Yaeger]

Alain,

Clever. Note that this only works if each group has the same number of records (in this case 4). If group 1 had 4 records and group 2 had 5 records, this wouldn't work. Do you have a trick for the case where the groups can have different numbers of records?

The reason you got the ICE027A with the INCLUDE statement, but NOT with the INCLUDE parameter of the OUTFIL statement is that the INCLUDE statement is processed before the INREC statement, whereas the OUTFIL statement is processed after the INREC statement. So the INCLUDE statement does not see the changes you make with the later INREC statement whereas the OUTFIL statement does see the changes you make with the earlier INREC statement.
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

[Alain Benveniste]

Frank,
Do you allow me to test C'AEXREC' and C'BDATA', if yes I will try to post something tomorrow.

Of course ...

Alain

[Frank Yaeger]

Alain,

A test of the starting string for the group (AEXREC) and the ending string for the group (******) would be more general. Is the BDATA test (record after end of group) really needed rather than the ****** test (record at end of group) for what you want to do?
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

[Alain Benveniste]

Frank,

In the 2nd example given by Hsubra we have duplicates (******---) records, one must be kept, one must be removed. That's why I wanted to use (BDATA). I understand what you mean by a more global solution and that's why I am asking you if you really want me to preserve this 'constraint' or can I bypass it ? I can also try to find something for the both cases.

Alain

[Frank Yaeger]

Is it just the unmatched ending record (******) that's the problem or any record in the group that has a duplicate outside the group?

I would consider the ideal group solution to be able to just keep the records from the starting string to the ending string for each group, and omit any records outside the groups. I would stipulate that every group has a starting and ending string. I would allow an ending string without a starting string and not consider that to be a group. I would allow any number of records before the first group and after the last group. I would allow duplicates anywhere. So the solution I'm asking about would be able to handle this:

Input
Y
X
Z
...
START
A
B
C
END
C
B
START
C
B
A
END
A
D
END
X
...

Output
START
A
B
C
END
START
C
B
A
END

Of course, I'm sure there are different variations of this in the real world that pose less difficulty for which solutions stipulating more constraints would be useful.
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

[hsubra]

Frank/Alain,

I am back from a small vacation. I saw the nice trick from Alain, when we have a number of records is same in each group. I have tried IFTHEN before my vaction , for me it did not work .. Nice trick Alain. As Frank told, I can not restrict the # of records in the group.

Because of the urgency, I am proposing HLL . But I want to try one more time with ICETOOL.

Is there any way i can reformat my input like this ..

INPUT:

[Alain Benveniste]

Hsubra,

The best way to solve the problem is to go step by step and not directly answer the request.
That

[hsubra]

Alain,

Thanks for the nice work. Give me sometime to understand the logic and let me try out .

Cheers,
Subra

[Alain Benveniste]

[Alain Benveniste]

I have only corrected minor mistakes in the examples
Here is the cntl statement for example I to IV that could be a little bit faster :